Você está na página 1de 83

1

Pulse Modulation
@ This material is copyright protected.
Disclaimer: This material is for the use of those students only who have registered for the course
EC-511: Digital Communication Systems in Autumn 2011-12 session.
2
Sampling Theorem:
The sampling theorem for strictly band-limited signals of finite
energy can be stated as below:
1. A band-limited signal of finite energy, which has no
frequency components higher than W Hz, is completely
described by specifying the values of the signal at instants of
time separated by 1/2Wseconds.
2. A band-limited signal of finite energy which has no frequency
components higher than W Hz, may be completely recovered
from a knowledge of its samples taken at the rate of 2W
samples per second.
3


=

=

n m
nf f nf G f mT t g ) ( ) ( ) (
0 0 0 0

Recall
(1)

=
=
m
T
mT t g t g ) ( ) (
0
0
: a periodic signal with a generating fn. g(t)
and period T
0
.
F.T. of a periodic signal with period T
0
consists of an infinite
sequence of delta functions occurring at integer multiples of the
fundamental frequency f
0
= 1/T
0
i.e., making a signal periodic in time domain has the effect of
sampling the spectrum of the signal in the frequency domain
i.e. (by the duality property of the F.T.) sampling a signal in the
time domain has the effect of making the spectrum of the signal
periodic in the frequency domain
4
g(t): an arbitrary signal of finite energy, which is
specified for all time
{g(nT
s
)}: Infinite sequence of samples obtained by
sampling g(t) at a uniform rate, once every T
s
sec.
g

(t): signal obtained by individually weighting the


elements of a periodic sequence of Dirac Delta
functions spaced T
s
sec. apart by the sequence of
numbers {g(nT
s
)}
Let
5
Analog Signal
Instantaneously
sampled version of
the signal
T
s
: sampling period, f
s
= 1/T
s
: sampling rate
6
The ideal sampled signal can be written as:
(2)
g

(t) has a mathematical form similar to that of the F.T. of a


periodic signal (eqn. 1)
Therefore, by applying the duality property of FT, the FT of g

(t):
(3)
By defn. of FT
[by taking the FT of both sides of (eqn. 1)]
(4)
7
Suppose the signal is strictly bandlimited,
i.e., G( f ) = 0 for f W
Suppose we choose the sampling period T
s
= 1/2W
Then the corresponding spectrum G

( f ) of the sampled signal


g

(t) will look like:


8
Spectrum of a strictly
band limited signal
g(t)
Spectrum of
sampled version of
g(t) for a sampling
period T
s
= 1/2W
9
Putting T
s
= 1/2W in eqn. (4),
(5)
From eqn. (3), the FT of g

(t) may also be expressed as:


(6)
Hence under the conditions:
eqn. (6) becomes:
(7)
10
Substituting eqn. (5) in eqn. (7), we have
(8)
If the sample values g(n/2W) of a signal g(t) are specified for
all time, then the FT G(f ) of the signal is uniquely determined.
In other words, the sequence {g(n/2W)} has all the information
contained in g(t). (Proof of 1
st
statement of Sampling Theorem)
Let us now consider the problem of reconstructing the signal g(t)
from the sequence of sample values {g(n/2W)}
11
Using eqn. (8)
(9)
(Proof of 2
nd
statement of Sampling Theorem)
12
The sampling rate of 2W samples per second, for a signal
bandwidth of W Hz, is called NYQUIST RATE, its reciprocal
(1/2W) (measured in seconds) is called NYQUIST INTERVAL
Sampling Theorem for Bandpass Signals:
Let for a signal m(t), its highest and lowest spectral components
are f
M
and f
L
.
For such a bandpass signal, the minimum sampling frequency
allowable is f
s
= 2(f
M
- f
L
), provided that either f
M
or f
L
is a
harmonic of f
s
.
If neither f
M
and f
L
is a harmonic of f
s
, a more general analysis is
required
13
Let B = f
M
f
L
and k f
M
/B
Then the sampling frequency required is:

N
k
B f
N
k
B f
s
s
2
and
1
1
2
in which k N, N: Harmonics number
14
Practical Difficulties in Signal Reconstruction
Signal sampled at Nyquist rate f
s
= 2W Hz The spectrum G

(f )
consists of repetition of G(f ) without any gap between successive
cycles.
Recover of g(t) requires ideal low-pass filter
1
st
Difficulty:
15
Solution:
Sample the signal at a rate higher than the Nyquist rate
This yields G

(f ) consisting of repetitions of G(f ) with a finite


gap (guard band) between successive cycles
Precaution:
Increase in sampling rate increase in width of the guard
band easing the problem of filtering
Increase in sampling rate extends the BW required for
transmitting the sampled signal
Accordingly an engineering compromise is called for
16
2
nd
Difficulty: The Treachery of Aliasing
The sampling theorem was proved on the assumption that
the signal g(t) is band-limited. All practical signals are time-
limited, i.e., they are of finite duration or width
It can be shown that a signal cannot be time-limited and
band-limited simultaneously. If a signal is time-limited, it
cannot be band-limited, and vice versa
That means all practical signals have infinite bandwidth, and
the spectrum G

(f ) consists of overlapping cycles of G(f)


repeating every f
s
Hz (the sampling frequency)
17
Recovered
Spectrum
Lost Tail
Lost tail gets
folded back
18
Because of infinite bandwidth, the spectral overlap is a constant
feature, regardless of the sampling rate
If the sampled signal is passed through an ideal LPF, the output is
not G(f) but a version of G(f) distorted as a result of two
separate causes
(i) The loss of tail of G(f) beyond | f |> f
s
/2 Hz
(ii) The reappearance of this tail inverted or folded onto the
spectrum
The frequency f
s
/2 where the spectra cross is called the FOLDING
FREQUENCY
19
The components of frequencies above f
s
/2 reappear as components
of frequencies below f
s
/2. This tail inversion is known as
SPECTRAL FOLDINGor ALIASING
A Solution: The Antialiasing Filter
The potential defectors are all the frequency components beyond
f
s
/2. We should eliminate these components (suppress) these
components from g(t) before sampling g(t)
This suppression of higher frequencies can be accomplished by an
ideal LPF of bandwidth f
s
/2 Hz. This filter is called the
ANTIALIASING FILTER
20
21
TDM and PAM
An important feature of sampling process is a conservation of
time, i.e., the transmission of the message samples engages the
communication channel for only a fraction of the sampling
interval on a periodic basis
We thereby obtain a TIME-DIVISION MULTIPLEX system,
which enables the joint utilization of a common communication
channel by a plurality of independent message sources without
mutual interference among them
22
23
The train of pulses corresponding to the samples of each signal are
modulated in amplitude in accordance with the signal itself.
Accordingly, the scheme of sampling is called PULSE
AMPLITUDE MODULATION (PAM)
24
Note 1: The use of TDM introduces a bandwidth expansion factor
N, because the scheme must squeeze N samples derived
from N independent message sources into a time slot equal
to one sampling interval
Note 2: The decommutator at the receiver should operate in
SYNCHRONISM with the commutator in the transmitter.
This synchronization is essential for a satisfactory operation
of the system
25
Natural Sampling:
Instantaneous sampling is hardly feasible.
Instantaneous samples at the transmitting end of the channel
have infinitesimal energy, and when transmitted through
a bandlimitedchannel give rise to signals having a peak
value which is infinitesimally small. Such infinitesimal
signals will inevitably be lost in background noise
A much more reasonable manner of sampling is referred to as
NATURAL SAMPLING
26
With natural sampling, as
with instantaneous sampling,
a signal sampled at the
Nyquist rate may be
reconstructed exactly by
passing the samples through
an ideal LPF with cutoff at
the frequency f
M
, where f
M
is
the highest frequency
spectral component of the
signal
27
Flat-top Sampling:
Pulses with tops contoured to follow the wave form of the signal,
are actually not frequently employed. Instead flat-topped pulses
are customarily used
Flat-top sampling has the merit that it simplifies the design of
the electronic circuitry used to perform the sampling operation
In sampling of this type the basebandsignal cannot be recovered
exactly by simply passing the samples through an ideal LPF.
However, the distortion need not be large
28
Flat-top Sampling:
29
Other Pulse Analog Modulation Techniques
PDM
PPM
30
The Quantization Process:
When quantizing a signal m(t) we create a new signal v(t) which is
an approximate to m(t)
The quantized signal v(t) has the great merit that it is, in large
measure, separable from the additive noise
The original continuous signal may be approximated by a
signal constructed of discrete amplitudes selected on a
minimum error basis from an available set
Example: 256 levels can be used to obtain the quality of
commercial color TV
31
Defn.: Amplitude quantization is defined as the process of
transforming the sample amplitude m(nT
s
) of a message signal m(t) at
a time t = nT
s
into a discrete amplitude v(nT
s
) taken from a finite set
of possible amplitudes
The signal amplitude m is specified by the index k if it lies inside
the interval
where L: total number of amplitude levels used in the quantizer
32
m
k
, k = 1, 2, , L: DECISION LEVELS /
DECISION THRESHOLDS
The quantizer output v equals v
k
if the input signal sample m
belongs to the interval I
k
i.e., v
k
represents all amplitudes of the
interval I
k
v
k
, k = 1, 2, , L: REPRESENTATION LEVELS /
RECONSTRUCTION LEVELS
The mapping v = g(m) is the quantizer characteristics, which is a
staircase function by definition
33
[The operation of quantization]
34
Quantized signal is separable from the additive noise. How?
Consider that the quantized signal has arrived at a repeater
somewhat attenuated and corrupted by noise.
Here the repeater consists of a quantizer and an amplifier
35
The probability of getting error can be reduced by increasing the
step size
However, increasing step size, results in an increased discrepancy
between the true signal m(t) and quantized signal v(t)
This difference m(t) v(t) can be regarded as noise and is called
QUANTIZATION NOISE
36
Quantization Noise
Let M: RV representing the quantizer input m
Let V: RV representing the quantizer output v
Q = M V: RV representing the quantization error q(= m - v)
The sample values m and v of RVs Mand V are related by
v = g(m)
Let the range of input m be (-m
max
, m
max
)
Assuming a uniform quantizer, the step size of the quantizer is
given by
where L: total number of representation levels
37
For a uniform quantizer, the quantization error Q will have its
sample values bounded by
-/2 q /2
If the step size is sufficiently small (i.e., the number of
representation levels L is sufficiently large), it is resonable to
assume that the quantization error Q is a UNIFORMLY
DISTRIBUTEDrandom variable
The pdf of the quantization error can thus be given by:
38
The variance or the mean-square value of the quantization error
Q
2
is given by

Typically, the L-ary number k, denoting the k


th
representation
level of the quantizer, is transmitted to the receiver in binary form
Let R: the number of bits per sample used in construction of the
binary code
39
Then

Hence
Let P: average power of the message signal m(t)
Then the output SNR of a uniform quantizer is given by:
40
Thus the output SNR of the quantizer increases
EXPONTIALLY with increasing number of bits per
sample, R
An increase in R requires a proportionate increase in the
channel (transmission) bandwidth
Thus the use of binary code for representation of message
signal provides an efficient method for the trade-off of
increased channel bandwidth for improved noise
performance
41
Pulse Code Modulation (PCM)
PCM is the most basic form of digital pulse modulation
In PCM, a message signal is represented by a sequence of coded
pulses, which is accomplished by representing the signal in
discrete form in both time and amplitude
The basic operations performed in the transmitter of a PCM
system are sampling, quantization and encoding
The basic operations in the receiver are regenerationof impaired
signals, decoding, and reconstruction of the train of quantized
samples
Regeneration also occurs at intermediate points along the
transmission path as necessary
42
43
Sampling:
1. Should be closely approximate to the instantaneous sampling
process
2. Sampling rate must be greater than twice the highest
frequency component Wof the message signal
3. The process facilitate us to use the TDM process
Quantization:
1. The process gives rise a new representation of the signal that
is discrete in both time and amplitude
44
Encoding:
45
1. The maximum advantage over the effects of noise in a
transmission medium is obtained by a using a binary code,
because a binary symbol withstands a relatively high level of
noise and easy to regenerate
2. There are several line codes that can be used for electrical
representation of binary symbols 1 and 0
(a) On-off signalling
(b) Nonreturn-to-zero signalling
(c) Return-to-zero signalling
(d) Bipolar return-to-zero signalling
(e) Split-phase or Manchester code
(f) Differential encoding
46
On-off signalling
Nonreturn-to-zero signalling
Return-to-zero signalling
Bipolar return-to-zero signalling
Split-phase or Manchester code
Differential encoding
47
Regeneration:
The equalizer shapes the received pulses so as to compensate for
the effects of amplitude and phase distortions produced by the
transmission characteristics of the channel
The timing circuitry provides a periodic pulse train, derived from
the received pulses, for sampling the equalized pulses at instants
of time where the SNR is a maximum
The samples so extracted is compared to a predetermined threshold
in the decision making device
48
PCM Implementation:
49
Question ( PCM): A CD records audio signals digitally by using
PCM. Assume that the audio signal bandwidth equals 15 kHz.
(a) If the Nyquist samples are uniformly quantized into L = 65536
levels and then binary-coded, determine the number of binary
digits required to encode a sample.
(b) If the audio signal has average power of 0.1 watt and peak
voltage of 1 volt. Find the resulting signal-to-quantization-noise
ratio (SQNR) of the uniform quantizer output in part (a)
(c) Determine the number of binary digits per second (bit/s) required
to encode the audio signal.
(d) For practical reasons, signals are sampled at a rate well above the
Nyquist rate. Practical CDs use 44100 samples per second. If L =
65536, determine the number of bits per second required to
encode the signal, and the minimum bandwidth required to
transmit the encoded signal.
50
Solution:
51
Non-uniform Quantization
The range of voltages covered by voice signals, from peak of
loud talk to the weak passages of weak talk is on the
order of 1000 to 1.
By using a non-uniform quantizer with the feature that the step
size increases as the separation from the origin of the
input-output amplitude characteristic is increased, the
large end step of the quantizer can take care of possible
excursions of the voice signal into the large amplitude
ranges that occur relatively infrequently
52
53
The use of a nonuniformquantizer is equivalent to passing the
baseband signal through a COMPRESSOR and then applying
compressed signal to a uniform quantizer
Among several choices, two compression laws have been
accepted as desirable standards by the CCITT
-law: used in North America and J apan
A-law: used in Europe and rest of the world and
international routes
54
-law:
: +veconstant
= 0 uniform quantization
i.e., -law is approximately linear at low input levels
corresponding to |m| << 1
and approximately logarithmic at high input levels
corresponding to |m| >> 1
= 255used in North America
55
A-law:
Thus the quantum of steps over the central linear segment, which
have the dominant effect on small signals, are diminished by the
factor A/(1+log A)
A value of A = 87.6 gives comparable results and has been
standardized by the CCITT
56
57
In order to restore the signal samples to their correct relativelevel,
we must, of course, use a device in the receiver with a
characteristic complementary to the compressor. Such a device is
called an EXPANDER
The combination of a compressor and an expander is called a
COMPANDER
58
[An input-output characteristic which provides compression]
59
Noise Considerations in PCM System:
1. Channel noise
2. Quantization noise
Channel Noise:
effect is to introduce bit error
need to minimize the average probability of symbol error
it is customary to model the channel noise originating at the
front end of the receiver as ADDITIVE, WHITE, and
GAUSSIAN
Can be made negligible by maintaining the SNR through the
provision of proper spacing between regenerative repeaters in the
PCM system
60
Quantization Noise:
it is essentially under the designers control
Can be made negligibly small through the use of adequate
number of representation levels in the quantizer and the
selection of a COMPANDER strategy matched to the
characteristics of the type of message signal being transmitted
61
Advantages of PCM system
1. Ruggedness to channel noise and interference
2. Efficient regeneration of the coded signal along the transmission
path
3. Efficient EXCHANGE of increase channel bandwidth for
improved SNR, obeying an expontial law
4. A uniform format for the transmission of different kinds of
basebandsignals, hence their integration with other forms of digital
data in a common network
5. Comparative ease with which message sources may be dropped
or inserted in a TDM process
6. Secure communication through the use of special modulation
schemes or ENCRYPTION
Increased system complexity and increased channel bandwidth
62
Differential Pulse Code Modulation (DPCM)
When a voice or video signal is sampled at a rate slightly higher
than the Nyquist rate, the resulting sampled signal is found to
exhibit a high correlation between adjacent samples
The meaning of this high correlation is that, in an average sense,
the signal does not change rapidly from one sample to the next,
with the result that the difference between adjacent samples has a
variance that is smaller than the variance of the signal itself
When these highly correlated samples are encoded, as in a
standard PCM system, the resulting encoding signal contains
REDUNDANT INFORMATION
By removing this redundancy before encoding we obtain a more
efficient coded signal
63
DPCM Transmitter
Let m(t): Basebandsignal
{m(nT
s
)}: Sequence of samples produced by sampling
the basebandsignal at a rate f
s
= 1/T
s
64
) (

s
nT m
(i) The input signal to the quantizer is
which is the difference between the unquantized input sample
m(nT
s
) and a prediction of it, denoted by
e(nT
s
): PREDICTION ERROR
(ii) This predicted value is produced by using a prediction filter,
whose input consists of a quantized version of the input sample
m(nT
s
)
65
A simple and yet effective approach to implement the prediction
filter is to use a tapped-delay-line filter, with the basic delay set
equal to the sampling period
[Prediction Filter]
66
(iii) By encoding the quantizer output, we obtain a variation of
PCM, which is known as DPCM. It is this encoded signal that
is used for transmission
(iv) The quantizer output may be expressed as
q(nT
s
): quantization error
The quantizer output e
q
(nT
s
) is added to the predicted value
to produce the prediction filter input
) (

s
nT m
which represents the quantized version of the signal m(nT
s
)
67
DPCM Receiver
The receiver consists of a decoder to reconstruct the quantized
error signal. The quantized version of the original input is
reconstructed from the decoder output using the same prediction
filter used in the transmitter
68
Delta Modulation (DM)
Procedure:
1. An incoming signal is oversampled(at a rate much higher than
the Nyquist rate)
2. The difference signal between the present sample value m(nT
s
) of
the input signal and the latest approximation to it, i.e., m(nT
s
) -
m
q
(nT
s
-T
s
) (= e(nT
s
)) is obtained
3. This difference signal is quantized into two levels, namely
corresponding to the +veand vedifferences respectively
4. In this way, DM provides a staircase approximation to the
oversampledversion of the signal
5. The quantizer output e
q
(nT
s
) (of e(nT
s
)) finally coded to produce
the desired DM signal
69
70
71
Implementation:
Let m(t): input signal
m
q
(t): staircase approximation
T
s
: Sampling period
e(nT
s
): Error signal representing the difference between
the present sample value m(nT
s
) of the input signal
and the latest approximation to it
e
q
(nT
s
):Quantizedversion of e(nT
s
)
72
DM Transmitter:
The comparator computes the difference between its inputs
The quantizer consists of a hard limiter with an input-output
relation that is a scaled version of the signumfunction
73
The quantizer output is then applied to an accumulator, producing
the result
DM Receiver:
74
Errors in DM:
Delta modulation is subject to two types of quantization errors:
1. Slope-overload error
2. Granular noise
75
1. m
q
(nT
s
) = m(nT
s
) + q(nT
s
) q(nT
s
): quantization error
Input sequence to the quantizer is
e(nT
s
) = m(nT
s
) - m(nT
s
-T
s
) - q(nT
s
-T
s
)
dt
t dm
T
s
) (
max

In order for the sequence of samples {m


q
(nT
s
)} to increase as
fast as the input sequence of samples {m(nT
s
)} in a region of
maximum slope of m(t), we require that the condition
be satisfied
76
2. Granular noise
In contrast to the slope-overload error, granular noise occurs
when the step size is too large relative to the local slope
characteristics of the input waveform m(t)
77
[Quantization error in delta modulation]
78
Coding speech at low bit rates
The use PCM at the standard rate of 64kbps demands a high
channel bandwidth for its transmission
In applications in which channel BW is at a premium, there is a
definite need for speech coding at low bit rates, while maintaining
acceptable fidelity or quality of reproduction
For coding speech at low bit rates, a waveform coder of
prescribed configuration is optimized by exploiting both
STATISTICAL CHARACTERIZATION OF SPEECH
WAVEFORMS and PROPERTIES OF HEARING
79
In particular the design philosophy has two aims in mind:
1. To remove redundancies from the speech signal as far
as possible
2. To assign the available bits to code the non-redundant
parts of the speech signal in a perpetually efficient
manner
Reduction of the number of bits per sample involves the
combined use of ADAPTIVE QUANTIZATION and
ADPATIVE PREDICTION
A digital coding scheme that uses both adaptive quantization
and adaptive prediction is called ADAPTIVE DIFFERENTIAL
PULSE-CODE MODULATION (ADPCM)
80
Adaptive quantization refers to a quantizer that operates with a
time-varying step-size (nT
s
), where, T
s
is the sampling period
81
The step size (nT
s
) is varied so as to match the variance
M
2
of the
input signal m(nT
s
). In particular
) (

) (
s M s
nT nT =
: a constant
: ) (

s M
nT
an estimate of the standard deviation
M
(nT
s
)
82
The use of adaptive prediction in ADPCM is justified, because
speech signals are inherently NONSTATIONARY
There are two schemes for adaptive prediction:
1. Adaptive prediction with forward estimation (APF), in
which unquantizedsamples of the input signal are used to derive
estimates of the predictor coefficients
2. Adaptive prediction with backward estimation (APB),
in which samples of the quantizer output and the prediction error
are used to derive estimates of the predictor coefficients
83
A signal processing scheme known as the least mean square
(LMS) algorithm for the predictor and an adaptive scheme for the
quantizer have been combined in a synchronous fashion for the
design of both the encoder and the decoder
The performance of this combination is so impressive at 32kbps
that ADPCM is now accepted internationally as a standard
coding technique for voice signals, along with 64 kbps using
standard PCM

Você também pode gostar